OCC-RAG: Optimal Cognitive Core for Faithful Question Answering

2026-05-30

AI Summary

This work addresses the limitations of current large language models in question answering, which rely heavily on parametric memory and often fail to ensure answer verifiability and multi-hop logical consistency. The authors propose OCC-RAG, a compact language model tailored for faithful question answering, trained on a novel synthetic data pipeline yielding three million samples that emphasize contextual faithfulness and calibrated abstention capability. A streamlined cognitive core architecture is introduced, enabling the model—during intermediate training stages—to generate structured multi-hop reasoning paths with literal citation-based provenance. Experimental results demonstrate that OCC-RAG (0.6B/1.7B parameters) substantially outperforms general-purpose models two to six times its size across multi-hop benchmarks including HotpotQA, MuSiQue, and TAT-QA, as well as faithfulness and abstention evaluations on ConFiQA and MuSiQue-Un.

Abstract

Recent progress in the development of language models has been defined by scale, with each generation absorbing more of the world's knowledge into its weights. However, many practical applications benefit more from robust reasoning than from extensive parametric knowledge. In this setting, task-specialized small language models (SLMs) offer a principled design choice. We introduce Optimal Cognitive Core (OCC), a family of SLMs built around this premise. As a variant of OCC, we present OCC-RAG, optimized for faithful question answering (QA) grounded in the provided context. This task directly aligns with the OCC design approach, requiring multi-hop reasoning over supplied passages while ignoring memorized knowledge. To train OCC-RAG, we implement a novel pipeline for synthesizing multi-context, multi-hop QA data at scale, producing a corpus of over three million examples targeting multi-hop reasoning, strict context faithfulness, and calibrated abstention. We release OCC-RAG-0.6B and OCC-RAG-1.7B, both mid-trained on this corpus. The models produce structured reasoning traces with source citations grounded in literal quotes from the context. Through OCC-RAG, we demonstrate that compact, task-specialized SLMs can match or exceed general-purpose models 2 -- 6x their size across multi-hop reasoning (HotpotQA, MuSiQue, TAT-QA), faithfulness (ConFiQA), and refusal (MuSiQue-Un) benchmarks.

Problem

• faithful question answering
• multi-hop reasoning
• small language models
• context faithfulness
• task-specialized models

Innovation

• Optimal Cognitive Core
• Small Language Models
• Faithful Question Answering
• Multi-hop Reasoning
• Retrieval-Augmented Generation